BACKGROUND
It is well-known that socio-economic status is related to participation in physical activity. Perceived barriers to involvement in physical activity play a key role in all major theories used to explain physical activity behavior. An understanding of the perceived barriers faced by different socio-economic segments of the population is important for the development of effective physical activity intervention programs and initiatives appropriate for the more disadvantaged. To date, very few studies have attempted to identify socio-economic differences in perceived barriers to physical activity and potential personal, social and environmental determinants of these differences.

OBJECTIVES
Adopting a multilevel ecological model of physical activity behavior, we examined socio-economic, psychosocial and environmental correlates of perceived barriers to participation in regular physical activity; whether any socio-economic differences in perceived barriers could be explained by psychosocial and environmental factors; and whether perceived barriers to physical activity could explain the observed relationships between indicators of socio-economic status (SES) and leisure-time and transport-related physical activity.

METHODS
This study used cross-sectional self-report data collected from the Physical Activity in Localities and Community Environments (PLACE) study in Adelaide, Australia. A stratified two-stage cluster sampling design was used to recruit 2650 English-speaking adults, aged 20-65, who were residents of private dwellings and able to walk without assistance. The study sample was drawn from residential addresses within 154 census collection districts (CCD) and classified based on their objective walkability characteristics and socio-economic status (SES) into four strata: high walkable/ high SES; low walkable/ high SES; high walkable/ low SES; and low walkable/ low SES. Area SES (CCD-level) was operationalized as the median weekly household income (positive indicator of SES) and the median household size (negative indicator of SES). Participants completed a questionnaire including information on socio-demographics, transport-related and leisure-time physical activity, perceived attributes of their local environment, and psychosocial correlates of physical activity behavior (enjoyment of, self-efficacy for, perceived benefit of, and barriers to physical activity). Generalized linear models with appropriate variance and link functions and with standard errors adjusted for clustering effects were used to examine the research questions. Factor analysis was used to identify latent dimensions of perceived barriers to physical activity.

RESULTS
Eight barrier-to-physical-activity factors were identified (weather, physical appearance, time, facilities/equipment, skill/knowledge, motivation, health, and social support). All barriers apart from lack of time were negatively related to either individual-level or area-level SES, or both. Lack of time was positively associated with individual-level household income. After controlling for psychosocial and environmental factors, SES indicators were significantly associated with all barriers except for lack of social support and lack of motivation. Environmental and psychosocial factors were independently related to all examined barriers. Psychosocial factors significantly accounted for socio-economic differences in all perceived barriers except lack of time. Perceived environmental factors partly explained socio-economic differences in the perceived barriers of lack of skill/knowledge, facilities/equipment and health. None of the examined barriers to physical activity reliably explained socio-economic differences in leisure-time and transport-related physical activity. However, the barriers of lack of motivation, social support, facilities/equipment, time and concerns about physical appearance were independently negatively associated with leisure-time physical activity (after controlling for all other explanatory variables). Interestingly, while a negative relationship was observed between lack of motivation and transport-related physical activity, positive relationships were observed between this type of physical activity and lack of skills/knowledge, lack of facilities/equipment and health concerns.

CONCLUSIONS
This study revealed substantial individual-level and area-level SES-differences in perceived barriers to physical activity. These differences were accounted for by psychosocial as well as perceived environmental factors, highlighting the need for multilevel educational and environmental intervention strategies to help participation in physical activity in the less affluent. The findings of this study also indicated that for those facing health and skill barriers, transport-related physical activity may be a more viable option to regular participation in physical activity.